Environmental reporting systems: the activity data behind every number you publish

The report sits downstream of data nobody currently collects

Plants generally know their electricity and gas consumption because they pay for it. Everything else is patchier: refrigerant top-ups recorded by a contractor on a service sheet, solvent usage inferred from purchases, waste quantities taken from a hauler's invoice rather than a weighing, water discharged and never measured. Scoping the report first and the data second is the usual sequence and the wrong one, because it produces commitments to figures nobody can source. Map what is actually recorded, by whom and how often, before agreeing what will be published.

Provenance is the part auditors examine

Each figure needs to carry where it came from, who provided it, which period it covers and when it entered the system. That sounds bureaucratic until a number is questioned and the only available answer is that it was in last year's file. Structured provenance also lets you correct a single source without recomputing everything by hand, and it exposes the uncomfortable cases where two figures rest on the same underlying record counted twice. Systems that accept a typed number with no evidence attached are storing an assertion, not a measurement.

Factors change, and history has to be reproducible

Converting activity data into reported quantities relies on published factors that are revised periodically. A recalculated history using new factors is not comparable with what was previously published, and an unrecalculated one is not internally consistent. The system has to record which factor version produced which figure, and the organisation has to decide and document its restatement policy before the question becomes urgent. This is one of the more common reasons a reported trend cannot be explained: the underlying activity barely moved and the factor set did.

Estimates, gaps and the discipline of saying so

Some data will not exist. A site without a water meter on a particular outfall, a contractor who reports by volume when you need mass, a period where a meter failed. Estimating is legitimate; concealing the estimate is not. Flag estimated values in the record, document the method, and report the proportion of the total that rests on estimation. The failure mode is quiet hardening: an estimate made once becomes a number in a file, gets copied forward, and within a couple of cycles is being cited as measured data by people who never saw the original caveat.

Waste and water are harder than energy

Energy arrives with an invoice and a meter. Waste arrives with mixed streams, a contractor's weighbridge you do not control, containers weighed full and estimated empty, and material that changes classification depending on how it is processed. Water involves abstraction, supply, internal reuse, treatment and discharge, and few plants measure all of them. Both areas depend on data supplied by third parties whose incentives are not aligned with your accuracy. Getting the reporting requirement into the waste contract at renewal is more effective than any amount of subsequent reconciliation.

Frequently asked questions

Can environmental reporting be done in a spreadsheet?

For a single site with few streams and no external assurance, yes, and pretending otherwise wastes money. The point at which it stops working is recognisable: several sites, several contributors, a requirement to show provenance, and a need to restate history when a factor changes. Spreadsheets handle calculation well and provenance, access control and version history badly, and it is those three that assurance work actually examines.

Who should own the environmental data set?

Ownership of the report belongs with the environmental function, but ownership of each input has to sit with whoever generates it: facilities for utilities, production for process consumption, procurement for purchased materials, the waste coordinator for disposal. Centralising collection without distributing responsibility produces one person chasing colleagues for numbers they treat as somebody else's priority, which is the most common reason reporting cycles overrun.

What do we do about supplier data we cannot verify?

Report it as supplier-provided and record what you did to test it, even if that was only a comparison against your own consumption records. Verification is a spectrum rather than a binary, and pretending unverified data is measured is worse than stating the limitation. Where a figure is material to the published result, put a data requirement into the contract at renewal, since suppliers respond to purchasing terms considerably faster than to requests from an environmental team.